Features
What is a model?
Learn what AI models do and how they fit into Dexto.A model is the AI that reads a request and decides what to say or do next. If an agent is an AI teammate, the model is the part that does the thinking.
Types of models
Section titled “Types of models”There are different types of models:
- Language models, including GPT, Claude, Gemini, and Grok. Examples include GPT-5 Mini, Claude Sonnet 4.6, Gemini 3.5 Flash, Grok 4.5, and Composer 2.5.
- Image generation models, such as GPT Image, Nano Banana, and Grok Imagine Image.
- Video generation models, such as Gemini Omni Flash Preview and Grok Imagine Video.
- Audio generation models, such as GPT Audio, Gemini TTS, and Lyria.
Dexto supports 400+ models, 5 video generation models, 2 image generation models, and 3 audio generation models, all through a unified credit system.
Model versus agent
Section titled “Model versus agent”Language models are the brain of an agent. Think of agents as AI teammates, and models as the brain that powers them.
Dexto also uses media generation models to create images, videos, and audio.
Choosing a model
Section titled “Choosing a model”Different models balance speed, cost, and capability differently. Some can also understand images, PDFs, audio, or video. Start with Dexto’s default model unless you have a specific reason to change it.
Dexto’s model picker shows a condensed view of the models you can choose from.
Check the models catalog for supported models and more details, and read our blog about choosing a model.
Bring your own key (BYOK)
Section titled “Bring your own key (BYOK)”You can bring your existing AI subscriptions and API keys into Dexto, so you can reuse your existing credits.
Currently supported providers:
| Provider | Connection options |
|---|---|
| OpenAI | API key or ChatGPT Login |
| Anthropic | API key |
| Cursor | API key |
Add or manage your connections in Model settings.